LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
7.2
Chapter 7
IPCC WG1 Fourth Assessment Report
The Carbon Cycle and the Climate System (P. Ciais, P. Cox, C. Heinze, S. Wofsy) (15 Pages)
7.2.1.X
Global budget interannual fluctuations [Ciais contribution to be modified by Steve when writing
this section]
The observed rate of atmospheric CO2 increase exhibits large variations every 2-4 years (fig ). This
observation cannot be accounted for by the small year-to-year variability (IAV) in global fossil CO2
emissions. This implies the existence of year-to-year fluctuations in the CO2 uptake by land and ocean
regions. Over the past two decades, low CO2 uptake rates uptake (i.e. high atmospheric growth rates)
occurred in 1987, 1994-95, 1997-98, and in 2002-2003. The record lowest uptake of 0.1 PgC only (highest
atmospheric growth rate of 2.7 ppm) occurred in 1998 during the strongest El Niño of the 20th century. The
yearly increase in atmospheric CO2 in 2002 and 2003 exceeded for the first time 2 ppm yr-1 in two
consecutive years (2.1 ppm yr-1 in 2002 and 2.6 ppm yr-1 in 2003), while there was only a rather weak El
Niño Conversely, high CO2 uptake (low growth rates) occurred in 1981, 1992-93 and 1996-97. Generally,
the high and low CO2 uptake periods correspond to the succession of La Niña and El Niño episodes, as has
long been recognized (Bacastow, 1976). There are at least two remarkable exceptions to this, the 1992-93
high uptake period popularly called the "post Pinatubo" period, and 2002-03 low uptake period where for the
first time since observations began, low uptake did not coincide with any developing El Niño.
Since the TAR, a consensus has emerged from top-down inversion studies that the CO2 fluxes IAV is mostly
due to land fluxes, and that the Tropical lands contribute strongly to the global terrestrial IAV (figure
Ciais4). Such a predominant terrestrial IAV forcing is indicated from: (1) CO2 concentrations observed at
different stations and used in atmospheric inversions by Bousquet et al. (2000), Roedenbeck et al. (2003) and
Baker et al. (2004); (2) consistent relationships between
zonally and (3) direct measurement of O2 and CO2 since 1991 (Manning, 2001).
7.2.1.3
7.2.1.3.1
Regional carbon fluxes inferred from bottom-up approaches [Ciais]
Terrestrial observations
Locally, stand-level biometric measurements can be used to estimate the carbon balance of land ecosystems
(Net Biome Productivity or NBP) but they are difficult to upscale since one must account for soil carbon
changes and for the impact of disturbances in controlling the long term averaged carbon balance ([Schulze, et
al., 2000] [Korner, 2003] ). In boreal forests for example, carbon losses caused by disturbances and
consumption of forest products comprises about 20% of NPP and can exceed the ecosystem carbon sink
[Shvidenko and Nilsson, 2003]. One cannot simply scale NBP from limited duration (less than 10 years)
eddy-covariance NEE data [Baldocchi, et al., 2001] because most eddy flux towers are located in young or
middle-aged stands and do not sample the effect of disturbances, and because the length of those records is
too short to average out the large fluctuations in NEE generally observed from one year to the next. The
alternative to measure stand NBP is to use chronosequences (e.g. [Mund, et al., 2002].
Regionally, extensive forest biomass inventories measurements estimate the measured volume growth of
stem wood and changes in forest area, the modeled fine litter and coarse debris production, corrected for
harvest and slash production in managed forests using bookkeeping type of models and other information for
upscaling. The advantage here is covering and understanding spatially variability, at the expense of detailed
knowledge of component processes. Systematic forest biomass inventories have been developed in temperate
and boreal forests in North America, Western Europe, Russia, China to measure the commercial value of
wood. In the tropics where there is no commercial forest inventory, stem growth measurements have been
carried out as part of research networks ([Malhi, et al., 2004; Phillips, et al., 1998];
http://www.geog.leeds/projects/rainfor/).
7.2.1.3.2
Terrestrial models
Do Not Cite or Quote
116099703
7-1
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Chapter 7
IPCC WG1 Fourth Assessment Report
Many spatially explicit models of terrestrial ecosystems have been developed to analyze and predict the
response of terrestrial carbon pools and fluxes to changing climate and atmospheric composition, and more
recently to land use and management practices (e.g. [McGuire, et al., 2001]). These models, similar to
those used in the TAR, encapsulate biogeochemical processes responsible for biomass production as driven
by climate and radiation, and calculate NPP as the difference between photosynthesis and autotrophic
respiration components. After mortality, litter and soil organic matter decomposition is calculated, allowing
estimates of the ecosystem carbon balance. The development of more realistic process-driven models enables
the handling of diverse resolutions and scales, and the use of assimilation methods to allow the efficient
processing of large amounts of data, particularly when using remote sensing observations (eg, [Wang and
McGregor, 2003]). The current generation of global terrestrial models are generally very crude in their
description of 1) forest growth and management, 2) carbon cycling over cultivated lands and wetland, 3) soil
organic matter decomposition, 4) frozen soil carbon dynamics, 5) the effects of land-use change and fire, 6)
coupling between carbon and nutrients cycling, 7) absorption of radiation by canopies, 8) effects of
biodiversity
1,60
average forest inventories (amazon)
NBP
1,40
models (amazon)
inversions NBP (amazon)
Stands (canif) (europe)
1,20
forest inventories (europe)
eddy flux + nee/nbp ratio (europe)
model average (europe)
1,00
inversion NBP (europe)
Min stand (pine central siberia)
0,80
forest inventories
b
model average NBP (siberia)
inversion NBP (siberia)
0,60
0,40
0,20
Fi
Mo
Inv
St
Fi
Mo
Ed
Inv
St
Fi
Inv
Mo
0,00
1
Figure 0. Estimates of the mean terrestrial carbon balance (in tC per ha per year) from different approaches
at various scales and for three distinct biomes, Green = Amazon ; Blue = Western Europe ; Red = Siberia
(data compiled by Ciais et al. 2005)
16
17
18
19
20
21
22
23
24
25
26
27
28
7.2.1.3.2
Attribution of bottom regional carbon fluxes to processes
It is fair to say that there are problems of quantifying NBP at various scales, and attributing it to the
controlling processes. Firstly, at the spatial scale of stands, NBP can only be defined from the in situ carbon
balance and thus does not include the fluxes of “displaced” carbon such as the losses to streams and rivers, or
the fate of wood products in harvested forests. A second issue deals with temporal scales, because NBP
contains “background” fluxes such as the formation of charcoal in fire-disturbed forests, the inclusion of
carbon into mineral soil horizons, or the entrainment of rock-weathered DIC to the ocean. Those fluxes are
part of the “background” natural carbon cycle and evolve on long time scales (see Figure XXX global carbon
cycle), whereas we are interested in NBP as the anomalous terrestrial uptake occurring on decadal time
scales in response to the current perturbation of the carbon cycle. It is virtually impossible to separate
“background natural” from “anthropogenic” NBP In field studies, except maybe when using the bombDo Not Cite or Quote
116099703
7-2
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Chapter 7
IPCC WG1 Fourth Assessment Report
radiocarbon signal to determine the soil component of ecosystem carbon storage (e.g. [Trumbore, et al.,
1996])
7.2.1.3.2 Ocean
Christoph ?
7.2.1.4
Regional carbon fluxes inferred from atmospheric CO2 observations and inverse modelling [Ciais]
Background
The atmosphere is a mixer and integrator of spatially and temporally varying surface fluxes, and so the
distribution and temporal evolution of CO2 in the atmosphere can be used to quantify surface fluxes, using
numerical models of atmospheric transport. This approach is known as inverse modelling. The atmospheric
approach to derive fluxes over the Earth surface from observations of concentration needs globally
distributed sites adapted to the specific circumstances at the level of continents or large ocean gyres, where
the precision and representativeness of the measurements should match that of the involved major source and
sink processes. Because of the long atmospheric life time of CO2, the horizontal gradients of concentrations
which carry the information on the magnitude and spatial distribution of sources and sinks, are quite small,
and remain difficult to detect with the present network of stations. In addition, atmospheric transport models
are not perfect, to a largely unknown extent, and so the inverted fluxes depend on the model used.
7.2.1.4.1
Atmospheric measurements
Atmospheric transport acts on varying fluxes at the surface to create CO2 concentration gradients. The
atmospheric transport is a fast process, with air from one hemisphere getting mixed globally on time scales
of roughly a year. Therefore, the gradients in atmospheric CO2 concentrations reflecting regional fluxes are
small compared to the background value of CO2. The global network of atmospheric CO2 concentration
measurement sites is shown in Figure 1 (see also [WMO/GAW, 2000]). Most of these sites are discrete
sampling sites, where flasks are collected approximately once a week, but there is a fewer number of in situ
continuous stations. The network is the sum of different national networks and programs, with
NOAA/CMDL contributing the largest dataset (see Figure 1). The atmospheric network is much denser over
the oceans than over the continents, and much denser in the Northern mid-to-high latitudes than in the
Tropics, or in the Southern high latitudes. Logically, the fluxes upwind of regions with higher coverage are
better constrained.
Figure 1. Global atmospheric network in 1989 and 2004 for all sites contributing to the GLOBALVIEW-CO2
data product. Discrete sampling means that flasks are collected each week or so, Continuous sampling
means that an instrument records CO2 in situ with a time step of less than an hour.
Do Not Cite or Quote
116099703
7-3
2/12/2016
Total pages: 13
LA1 Extended Draft
Chapter 7
IPCC WG1 Fourth Assessment Report
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Gradients in CO2 concentrations have long been characterized as a function of latitude ([Keeling, et al.,
1989]; [Tans, et al., 1990]). There is an inter-hemispheric CO2 difference, which has been increasing
proportionally with the fossil CO2 emissions in the Northern hemisphere during the past 40 years. The
observed average annual mean CO2 difference between the Northern Hemisphere and the Southern
Hemisphere equals 3 ppm during the period 1991-2000, and 4 ppm during 2001-2005. There are also
gradients of CO2 in longitude, which reflect the activity of regional fluxes. But the atmospheric transport in
longitude is very fast at mid and high latitudes so that CO2 gradients in longitude are generally smaller than
those in latitude. For instance Fan et al. 1998 inferred a large sink over North America from a CO2 gradient
of -0.4 ppm between Atlantic and Pacific stations. There are also temporal CO2 variations, with a seasonal
cycle amplitude of up to 15 ppm in the Northern Hemisphere, being mostly implied by the photosynthesis
and respiration processes, with a small contribution of fossil fuel emissions. The gradients in CO2
concentration also exhibit important year-to-year variations. For instance the inter-hemispheric CO2
difference varied from one year to the next between 2 and 5 ppm during 1991-2000.
Do Not Cite or Quote
116099703
7-4
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
7.2.1.4.2
Chapter 7
IPCC WG1 Fourth Assessment Report
Atmospheric inversion models
Using as an input the atmospheric CO2 data described above, the atmospheric inverse models determine a
global distribution of surface CO2 sources and sinks which minimize a mismatch between modelled and
observed gradients of concentrations, accounting for measurements errors. Bayesian synthesis inversions, the
most widely used method since pioneered by [Enting, et al., 1995]are required simultaneously to fit
optimally the available data without deviating too much from an assigned a priori flux model. When doing
so, a number of modes of variability of the fluxes are explicitly solved for utilizing the information contained
in the atmospheric observations, while a number of other modes are constrained by the a priori settings. One
can explicitly propagate errors on atmospheric measurements and errors on the prior fluxes onto the sought
fluxes, thus delivering a very useful set of uncertainty estimates. Generally, Gaussian random errors are
assumed for the measurements and for the a priori fluxes, translating into Gaussian errors on the inverted
fluxes. We report this random uncertainty as colored bars in the inversion fluxes of Figure 2.
In addition, inverse model results are biased by a certain choice of arbitrary settings. Those biases are
difficult to quantify, and even harder to relate to the unknown truth. Arbitrary settings which can bias
inverted fluxes include 1) the choice of a particular atmospheric transport model, 2) the assumed ability for
this model to match a point observation with a grid-box model simulation (representation error), 3) the
choice of fixed a priori temporal flux variations (e.g. as in [Gurney, et al., 2002; Gurney, et al., 2003]) or
spatial flux variation within a given region (see discussion in [Kaminski, et al., 2001]). There is also a bias
arising from the fact that fossil fuel emissions are generally considered to be of perfectly known magnitude
and patterns, so that their effect can be easily modelled and subtracted from atmospheric CO2 data to solve
for “residual” land and ocean fluxes. These various biases translate into a spread in the inverted fluxes,
which can be partly understood by performing an ensemble of different sensitivity inversions where the
arbitrary settings are varied (e.g. using different transport models through the same inversion procedure, as
done by the TRANSCOM-3 project ref). We report the spread of mean fluxes deduced from inversion
ensembles with different settings as a grey bar in Figure 2.
The fluxes inferred from inverse models also depend critically on how the atmospheric observations are
treated. Inhomogeneities and gaps in the data are inevitable, and the increasing number of sites in the
network (Figure 1) generates spurious changing estimates of regional fluxes as an artefact of adding stations
downwind of regions previously not well constrained. Further, network inhomogeneities, coverage denser
over oceans [Patra, et al., 2005], and calibration is also a source of bias on regional fluxes inferred from
inversions. Several studies showed that the network is not dense enough for stations to be redundant. That
means that in some cases, adding or removing sites ([Bousquet, et al., 1999]; [Law, et al., 2003]), or varying
errors on the data [Krakauer, et al., 2004] have strong implications for the inverted fluxes.
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Figure 2. Ocean and land fluxes from inversions summed into large latitude bands, and over the globe during the period
1992-1996. Blue = Ocean fluxes, Green = land fluxes, Yellow = total land plus ocean fluxes obtained after subtracting
Do Not Cite or Quote
116099703
7-5
2/12/2016
Total pages: 13
LA1 Extended Draft
Chapter 7
IPCC WG1 Fourth Assessment Report
the non-optimized fossil fuel emissions in red. The mean flux from different ensembles of inversions is reported
together with random errrors and the range of bias due to different settings within each ensemble. Error boxes in colour
are the average of 1-s Gaussian random errors returned by each member of the ensemble. Grey error boxes are the
spread of the mean fluxes from different arbitrary settings. TAR = range of mean fluxes from Third Assessment Report
; G02 = [Gurney, et al., 2002] ensemble using annual mean CO2 observations with grey error from 16 transport models
; G03 is the [Gurney, et al., 2003] ensemble using smoothed monthly CO2 observations with grey error from 13
transport models ; P04 is the [Peylin, et al., 2005a] ensemble with grey error from 3 transport models times 3 grouping
of large regions times 3 inversion set of regions ; R03 is the [Rodenbeck, et al., 2003] ensemble where the fluxes are
solved for many small regions using monthly flask data, with the grey error from their different sensitivity inversions
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
7.2.1.4.3
Atmospheric inversion results: mean regional fluxes
In general, the confidence on the long-term mean fluxes is lower than on the year-to-year variations of the
fluxes. Many inversions ensembles, each with different settings, have been produced since the TAR (Baker
et al., 2004; [Gurney, et al., 2002; Gurney, et al., 2003; Peylin, et al., 2002] [Peylin, et al., 2005a;
Rodenbeck, et al., 2003]. An updated breakdown of oceans and land fluxes for large latitude bands is given
in Figure 2. Inversion results are reported with their random estimation errors and a necessarily imperfect
estimate of the (range of) bias implied by various settings in each ensemble. In the TRANSCOM-3 project
were compared 16 different transport models through the same inverse procedure [Gurney, et al., 2002;
Gurney, et al., 2003]. The results showed that differences between transport models cause uncertainties of
smaller (or similar) magnitude than the random estimation error. This means that, for most regions, adding
new measurements would be most beneficial for inversions. Over northern land regions, (e.g. boreal Asia),
the models differ strongly in the way they respond to the seasonal cycle of the terrestrial CO2 exchange, so
that improving models is needed, Tropical regions, in particular Tropical lands, could not be adequately
constrained because of insufficient data, a general feature of all inversions (see Figure 2).
All possible inversion settings were certainly not explored in the TRANSCOM-3 ensembles. In particular the
effect of the 'a priori aggregation' bias resulting from setting a fixed spatial pattern to the fluxes within each
region. Inverting the fluxes at a much higher spatial resolution, up to the model grid itself [Kaminski, et al.,
1999] [Rodenbeck, et al., 2003] [Peylin, et al., 2005b]) overcomes that limitation, but it implies nevertheless
that the a priori flux model still influences the inversion result. Figure 2 shows differences above their 1-
errors between TRANSCOM-3 and [Rodenbeck, et al., 2003] ensembles over Tropical and northern
continents for the period 1992-1996. Part of these differences can be explained by different methodologies,
in particular a different network of 35 stations (monthly flask data) in [Rodenbeck, et al., 2003] vs. 76
stations (smoothed flask data) in TRANSCOM-3. Figure 3 gives a range of 0.4-2.3 PgC y-1 for uptake in
Northern continents, and a range of 0.4 PgC y-1 sink to 1.8 pgC y-1 source over tropical continents during
1992-96. Compensation effects obviously exist between the inversion results in the North and in the Tropics.
Figure 3 gives the breakdown of inversion fluxes for five large regions of the Northern Hemisphere: North
America, North Atlantic, Europe, North Asia, and North Pacific. There are important differences in the mean
of different ensembles, except over North America where all inversion fluxes generally agree within their
errors to a mean sink ranging between 0.6 and 1.1 PgCy-1, although large uncertainties still pertain to that
estimate (full range 0 to 1.6 PgC y-1). Larger differences between the inversion ensembles are found over
Europe (range -0.9 to +0.2 PgC y-1) and over North Asia (range -1.2 to +0.3 PgC y-1), a latter region where
inversion are sensitive to transport model choices, as seen from the large grey error in Figure 3
Do Not Cite or Quote
116099703
7-6
2/12/2016
Total pages: 13
LA1 Extended Draft
Chapter 7
IPCC WG1 Fourth Assessment Report
QuickTime™ and a
TIFF (LZW) decompressor
are needed to see this picture.
Figure 3. Blue, Green = Ocean and land fluxes from inversions in the Northern Hemisphere. Bottom-up continentalwide or ocean-basin-wide estimates are given separately. Magenta = Forest biomass inventory data analyzed by
[Nabuurs, et al., 1997] (Europe) and [Shvidenko and Nilsson, 2003] (Siberia) ; Yellow = Forest biomass inventory data
analyzed by [Goodale, et al., 2002], including Russia, Mongolia and China for N. Asia ; Orange = bottom-up synthesis
of the net carbon balance of coterminous U.S. [Pacala, et al., 2001], Geographic Europe [Janssens, et al., 2003], and
Russia [Shvidenko and Nilsson, 2003]. Cyan = air sea basin scale fluxes from [Takahashi, et al., 1999]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
What is robust, i.e. inversion ensembles agree within their errors ?
- The ocean uptake deduced from all inversions is higher in the Northern Extratropics than in the Southern
Extratropics.
- The tropical oceans are outgassing CO2 to the atmosphere (range 0 to 1.5 PgC y-1), as expected from
upwelling of CO2-rich waters [Feely, et al., 1999]
- The global land vs. ocean flux partitioning from inversions is within the uncertainty of the O2 derived
global budgets. This result cannot serve as an independent verification since some inversions precisely
use that constraint.
- The Northern Hemisphere land fluxes are better constrained than the Tropical land fluxes
- There is a mean terrestrial sink in the Northern Hemisphere, with a fraction in North America, in Europe
and Northern Asia. Very large uncertainties pertain to each regional flux estimate, but the smallest spread
is found over North America
- The North American carbon sink estimated by recent inversions is on average lower than an earlier,
widely cited study [Fan, et al., 1998], but this latter estimate remains within the inversion range of
uncertainty.
What is not robust, i.e. inversion ensembles disagree within their errors ?
- The partition of terrestrial fluxes between northern and tropical lands.
- The sink function of tropical lands : if tropical lands inverted fluxes give a CO2 source of similar
magnitude than tropical deforestation in Table XX Steve deforestation table , then tropical undisturbed
terrestrial ecosystems are about carbon neutral ; If tropical lands inverted fluxes are a net sink of CO2,
then undisturbed tropical ecosystems are a huge sink of CO2
Do Not Cite or Quote
116099703
7-7
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Chapter 7
IPCC WG1 Fourth Assessment Report
- The mean flux estimate of most individual regions is not robust through inversions
7.2.1.4.4
Mean regional fluxes, bottom-up estimates and inversions
Bottom up estimates of regional carbon fluxes have been summarized in section 7.2.1.4.4. Comparing
inversion estimates of large-scale sources and sinks with bottom up estimates is not an easy task because 1)
inversion fluxes may already contain a certain amount of a priori knowledge of bottom-up fluxes, especially
of air-sea fluxes, so that the two approaches are not fully independent 2) the time period for which inversion
models and bottom-up estimates are available is often not consistent, in presence of fluxes that can vary
substantially from one year to the next (see 7.2.1.4.5), 3) the distribution of CO2 fluxes given by inversions
does not match the distribution of carbon storage generally given by bottom up , in presence of lateral fluxes.
Lateral fluxes act to transport carbon away from where CO2 is exchanged to or from the atmosphere, in such
a manner that the regional changes in carbon storage have a geographically distinct distribution than the
CO2 fluxes [Tans, et al., 1995] [Sarmiento and Sundquist, 1992]. Bottom up methods are measuring carbon
storage changes, and inversions CO2 fluxes. Lateral fluxes include the emissions of reduced carbon
compounds (VOCs, CO, CH4) by ecosystems and anthropogenic combustion processes, transported and
oxidized by OH chemistry [Enting and Mansbridge, 1991] [Folberth, et al., 2005] [Suntharalingam, et al.,
2005], trade of wood and food products harvested from ecosystems [Ciais, et al., 2005; Imhoff, et al., 2004],
and riverine transport of dissolved inorganic and organic carbon to the ocean (Figure XX Global C Cycle)
[Aumont, et al., 2001] [Meybeck, 1987]. The main results from comparison of bottom-up with inversion
mean fluxes are
-
Inversions determine a higher ocean sink in the Northern Extratropics than in the Southern
Extratropics, in contrast with oceanographic estimates of [Takahashi, et al., 2002; Takahashi, et al.,
1999], although new measurements of pCO2 in winter over the Southern Ocean may reconcile both
estimates [Metzl, et al., 1999].
-
Inversions tend to estimate higher terrestrial carbon uptake over Northern Hemisphere lands than
bottom up estimates [Nabuurs, et al., 1997] [Shvidenko and Nilsson, 2003] [Goodale, et al., 2002]
[Pacala, et al., 2001] [Janssens, et al., 2003] [Shvidenko and Nilsson, 2003] (see Figure 3), but part
of this discrepancy can be explained by laterally transported carbon via reduced carbon compounds,
rivers and food and wood trade
-
Inversions determine that tropical land regions are either close to neutral, or a sink of CO2,
qualitatively in agreement with limited forest biomass changes measurements in the Amazon
[Phillips, et al., 1998] [Malhi and Grace, 2000]
7.2.1.4.5
Atmospheric inversion results: interannual changes in regional fluxes.
Three inversion ensembles calculated the changes in monthly regional fluxes over the last two decades
([Bousquet, et al., 2000] [Rodenbeck, et al., 2003]; [Baker, et al., 2005]). As for mean regional fluxes, these
studies report a random error and the error range of sensitivity inversions with different settings (set of
atmospheric stations, atmospheric measurement errors, transport model, the choice, a priori flux constraints,
spatial or temporal scales at which the fluxes are solved for).
[Bousquet, et al., 2000] used large regions and varied different inversion settings for the period 1980-1998.
They found that the year-to-year regional flux changes could be more robustly inverted than their mean
values.
[Rodenbeck, et al., 2003] used one transport model and an inversion at the resolution of model grid for the
period 1982-2002, and changed different inversion settings. In particular, different stations from the CMDL
network (Figure 1) were used, from 11 sites up 35 sites, showing an impact on the inversion mean fluxes but
less on broad regions variability. The effect of using interannually varying winds was concluded to have a
relatively small impact on the fluxes variations.
Do Not Cite or Quote
116099703
7-8
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Chapter 7
IPCC WG1 Fourth Assessment Report
[Baker, et al., 2005] used large regions but 13 different transport models for the period 1988-2002. Their
results suggest that the spread of transport models has a lesser impact on the year-to-year regional flux
anomalies than on the mean fluxes (see 7.2.1.3.3). In other words, the changes of regional fluxes owe more
to errors in atmospheric data, a priori constraints, than to inter-model differences.
What is robust, i.e. inversion ensembles agree within their errors ?
- At the scale of the globe, the IAV of land fluxes is larger than the one of ocean fluxes.
- At the scale of large latitude bands (Figure 1) the IAV of land fluxes is larger than the one of ocean fluxes
- Tropical land fluxes exhibit on average a larger variability than temperate and boreal land fluxes, and
their variability suggests unambiguously an El Niño forcing, with anomalous sources during El Niño
episodes in 1982-1983, 1986-1987 and 1997-1998, and anomalous sinks during La Niña episodes.
- The 1997-1998 high CO2 growth rate episode is attributed to an abnormal source in the Tropics, with a
significant fraction of it in South East Asia, in good agreement with bottom up estimates (fires in
Indonesian moist forests)
- The Tropical Pacific ocean is a region where the significance of the inversion results for year-to-year
changes is high, in good agreement with bottom up estimates (pCO2 observations)
What is not robust, i.e. inversion ensembles disagree within their errors ?
- The 1992-1993 low CO2 growth rate episode during the post Pinatubo climate disturbance is attributed
either to the Northern mid latitudes by Bousquet et al. and to the Tropical lands by the two other
ensembles
- At the scale of continents or ocean basins in the Northern Hemisphere, the errors increase and the
significance of the inverted flux inter-annual changes is quickly lost.
Figure 4. Ocean and land fluxes year-to-year variations from inversions summed into large latitude bands,
and over the globe during approx. the past 20 years. Three different inversion ensembles are shown, and for
each flux and each region, the long term mean values has been substracted and the seasonal signal has been
removed
Do Not Cite or Quote
116099703
7-9
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
7.2.1.4.6
Chapter 7
IPCC WG1 Fourth Assessment Report
Year to year changes in regional fluxes, inversion and bottom-up
- Globally, inversions results agree with independent ocean carbon model calculations on the rather small
variability of air sea fluxes (± 0.5 Pg y-1 between extremes), and conversely on to large variability of
land fluxes (± 4 PgC y-1 between extremes). Yet, inversions and ocean models differ on the geographic
contributions to the year-to-year variability.
- In the Northern Extratropics (Figure 3), Baker et al. and Bousquet et al. obtain a significant variability in
air sea fluxes, in contrast to [Rodenbeck, et al., 2003]. don’t. On the one hand, over the North-Atlantic
ocean, Gruber et al. extrapolated a large variability comparable to Bousquet et al. (extremes of ± 0.3 PgC
y-1) using data from the Bermuda station during 1984-2000. On the other hand, for that region[McKinley,
et al., 2004] modelled a small variability (extremes of ±0.1 PgC y-1) comparable to the one of Rödenbeck
et al 2003, and suggested that a larger variability could be an effect of the large region settings (ref). Note
that all inversions are not strictly independent from bottom-up studies, as they include a priori
information to constrain the air-sea flux variations, which may in turn influence their results.
- El Niño and La Niña episodes force respectively a large release and uptake of CO2 over the tropical
continents, in particular over the Amazon. Fluxes anomalies are large (2-3 Gt y-1) and in general good
agreement with many bottom-up biospheric model calculations. In the Equatorial Pacific ocean, all
inversions obtain a flux variability which compares well in magnitude and timing with ocean models
results [Le Quere, et al., 2003; Le Quere, et al., 2000; McKinley, et al., 2004] [McKinley, et al., 2003] and
with pCO2 observations (Feely et al. 2002). The flux variability in that region is paced by El Niño, with
a smaller net CO2 source to the when upwelling of CO2-rich waters diminishes.
- - The 1992-1993 low CO2 growth rate episode during the post Pinatubo climate disturbance is not a
robust result of inversions. [Bousquet, et al., 2000] find an enhanced terrestrial uptake in the Northern
Hemisphere, while Baker et al. and Rodenbeck et al. place it in the Tropics. Analysis of remotely-sensed
vegetation activity and a terrestrial carbon model calculation by [Lucht, et al., 2002] produces good
agreement with Bousquet et al. Apart from higher NPP, lower respiration or a combination of both
conventionally expected from the cooling, an overall increase in photosynthesis could have occurred via
the increased fraction of diffuse sunlight [Roderick, et al., 2001]. But a clear causal relationship between
changes in climate or in radiation after the Pinatubo eruption and regional CO2 fluxes anomalies has not
been found yet.
- The 1997-1998 high CO2 growth rate period can be consistently explained by abnormally high and
intense fire emissions. For Indonesia alone, a source of + 0.8 to +2.6 GtC from peat forest fires was
estimated by [Page, et al., 2002]. For the globe, approximately 2/3 of the 1997-1998 growth rate excess
was attributed to fire emission anomalies, with main contributors South East Asia (60%), South America
(30%) and a small Siberian contribution [van der Werf, et al., 2004]. Similarly, [Langenfelds, et al., 2002]
analyzed the correlations in the interannual growth rate of CO2 and other species at 10 stations and linked
the 1997-1998 (and the 1994-95) anomalies and to high fires emissions as a single process. The
relationship between El Niño and CO2 emissions from fires is however not uniform, with for instance in
Africa drier conditions determining less fire emissions due to less biomass (ref). In addition, co-varying
processes such as reduced NPP (ref) caused by the dry conditions over tropical forests during El Niño
episodes may be superimposed on fire emissions.
Aumont, O., et al. (2001), Riverine-driven interhemispheric transport of carbon, Glob. Biogeochem. Cycle,
15, 393-405.
Baker, D. B., et al. (2005), TransCom 3 inversion intercomparison: impact of transport model errors on the
interannual variability of regional CO2 fluxes, 1988-2003, Journal of Geophysical Research, submitted.
Baldocchi, D., et al. (2001), FLUXNET: A New Tool to Study the Temporal and Spatial Variability of
Ecosystem-Scale Carbon Dioxide, Water Vapor, and Energy Flux Densities, Bulletin of the American
Meteorological Society, 82, 2415-2434.
Do Not Cite or Quote
116099703
7-10
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Chapter 7
IPCC WG1 Fourth Assessment Report
Bousquet, P., et al. (2000), Regional Changes in Carbon Dioxide Fluxes of Land and Oceans Since 1980,
Science, 290, 1342-1346.
Bousquet, P., et al. (1999), Inverse modelling of annual atmospheric CO2 sources and sinks 2. Sensitivity
study, Journal of Geophysical Research, 104, 26179-26193.
Ciais, P., et al. (2005), On the Horizontal Displacement of Carbon Associated to Agriculture and How it
Impacts Atmospheric CO2.
, Global Biogeo. Cycles, submitted.
Enting, I. G., and J. V. Mansbridge (1991), Latitudinal Distribution of Sources and Sinks of Co2 - Results of
an Inversion Study, Tellus Series B-Chemical and Physical Meteorology, 43, 156-170.
Enting, I. G., et al. (1995), A synthesis inversion of the concentration and d13C of atmospheric CO2., Tellus,
47B, 35-52.
Fan, S., et al. (1998), A Large Terrestrial Carbon Sink in North America Implied by Atmospheric and
Oceanic Carbon Dioxide Data and Models, Science, 282, 4.
Feely, R. A., et al. (1999), Influence of El Nino on the equatorial Pacific contribution to atmospheric CO2
accumulation, Nature, 398, 597-601.
Folberth, G., et al. (2005), On the role of atmospheric chemistry in the global CO2 budget
, Geophys Res Letters.
Goodale, C. L., et al. (2002), Forest carbon sinks in the Northern Hemisphere, Ecological Applications, 12,
891-899.
Gurney, K. R., et al. (2002), Towards robust regional estimates of CO2 sources and sinks using atmospheric
transport models, Nature, 415, 626-630.
Gurney, K. R., et al. (2003), TransCom 3 CO2 inversion intercomparison: 1. Annual mean control results
and sensitivity to transport and prior flux information, Tellus B, 55, 555-579.
Imhoff, M. L., et al. (2004), Global patterns in human consumption of net primary production, 429, 870-873.
Janssens, I. A., et al. (2003), Europe's terrestrial biosphere absorbs 7 to 12% of European Anthropogenic
emissions, Science, 300, 1538-1542.
Kaminski, T., et al. (1999), A coarse grid three-dimensional global inverse model of the atmospheric
transport. 2. Inversion of the transport of CO2 in the 1980s, Journal of Geophysical Research, press.
Kaminski, T., et al. (2001), On aggregation errors in atmospheric transport inversions, Journal of
Geophysical Research-Atmospheres, 106, 4703-4715.
Keeling, C. D., et al. (1989), A three-dimensional model of atmospheric CO2 transport based on observed
winds, 1, Analysis of observational data, in Aspects of Climate Variability in the Pacific and Western
Americas, Geophysical Monogr. Ser., edited by D. H. Peterson, pp. 165-236, AGU, Washington, D.C.
Korner, C. (2003), Slow in, Rapid out--Carbon Flux Studies and Kyoto Targets, Science, 300, 1242-1243.
Krakauer, N. Y., et al. (2004), Using general crosss-validation to select parameters in inversions for regional
carbon fluxes, Geophys Res Letters, 31.
Langenfelds, R., et al. (2002), Interannual growth rate variations of atmospheric CO2 and its delta C-13, H2, CH4, and CO between 1992 and 1999 linked to biomass burning, Glob. Biogeochem. Cycle, 16, 10481048.
Law, R. M., et al. (2003), Data and modelling requirements for CO2 inversions using high-frequency data,
Tellus Series B-Chemical and Physical Meteorology, 55, 512-521.
Le Quere, C., et al. (2003), Two decades of ocean CO2 sink and variability, Tellus Series B-Chemical and
Physical Meteorology, 55, 649-656.
Le Quere, C., et al. (2000), Interannual variability of the oceanic sink of CO2 from 1979 through 1997, Glob.
Biogeochem. Cycle, 14, 1247-1265.
Lucht, W., et al. (2002), Climatic control of the high-latitude vegetation greening trend and Pinatubo effect,
Science, 296, 1687-1689.
Malhi, Y., et al. (2004), The above-ground wood productivity and net primary productivity of 100
neotropical forests, Global Change Biology.
Malhi, Y., and J. Grace (2000), Tropical forests and atmospheric carbon dioxide, Trends in Ecology and
Evolution, 15, 332-337.
McGuire, A. D., et al. (2001), Carbon balance of the terrestrial biosphere in the twentieth century: Analyses
of CO2, climate and land use effects with four process-based ecosystem models, Glob. Biogeochem. Cycle,
15, 183-206.
McKinley, G. A., et al. (2004), Mechanisms of air-sea CO2 flux variability in the equatorial Pacific and the
North Atlantic, Global Biogeochem. Cycles, 18.
Do Not Cite or Quote
116099703
7-11
2/12/2016
Total pages: 13
LA1 Extended Draft
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Chapter 7
IPCC WG1 Fourth Assessment Report
McKinley, G. A., et al. (2003), Interannual variability of air-sea O-2 fluxes and the determination of CO2
sinks using atmospheric O-2/N-2, Geophys. Res. Lett., 30.
Metzl, N., et al. (1999), The annual fCO2 cycle and the air-sea CO2 flux in the sub-Antarctic Ocean, Tellus.
Meybeck, M. (1987), American Journal of Science, 287, 401-428.
Mund, M., et al. (2002), Growth and carbon stocks of a spruce forest chronosequence in central Europe.,
Forest Ecology and Management
, 171.
Nabuurs, G. J., et al. (1997), The role of european forests in the global carbon cycle - a review, Biomass and
Bioenergy, 13, 345-358.
Pacala, S. W., et al. (2001), Consistent Land- and Atmosphere-Based U.S. Carbon Sink Estimates, Science,
292, 2316-2320.
Page, S. E., et al. (2002), The amount of carbon released from peat and forest fires in Indonesia during 1997,
Nature, 420, 61-65.
Patra, P. K., et al. (2005), Dependence of CO2 flux determination on observational networks in inverse
modelling of atmospheric CO2, Journal of Geophysical Research.
Peylin, P., et al. (2002), Influence of transport uncertainty on annual mean and seasonal inversions of
atmospheric CO2 data, Journal of Geophysical Research-Atmospheres, 107.
Peylin, P., et al. (2005a), Multiple constraints on regional CO2 fluxes variations over land and oceans,
Global Biogeo. Cycles, press.
Peylin, P., et al. (2005b), Daily CO2 flux estimates over Europe from continuous atmospheric measurements:
1. Inverse methodology, Atmospheric Chemistry and Physics Discussions, 5, 1647-1678.
Phillips, O. L., et al. (1998), Changes in the carbon balance of tropical forest: evidence from long-term plots,
Science, 282, 439-442.
Rodenbeck, C., et al. (2003), CO2 flux history 1982-2001 inferred from atmospheric data using a global
inversion of atmospheric transport, Atmospheric Chemistry and Physics, 3, 1919-1964.
Roderick, M. L., et al. (2001), On the direct effect of clouds and atmospheric particles on the productivity
and structure of vegetation, Oecologia, 129, 21-30.
Sarmiento, J. L., and E. T. Sundquist (1992), Revised budget for the oceanic uptake of anthropogenic carbon
dioxide, Nature, 356, 589-593.
Schulze, E. D., et al. (2000), Interactions between the carbon and nitrogen cycles and the role of biodiversity:
A synopsis of a study along a north - south transect through Europe, in Carbon and nitrogen cycling in
European forest ecosystems. 468 -491, Ecological Studies Vol 142, edited by E.-D. Schulze, Springer
Verlag, Heidelberg.
Shvidenko, A. Z., and S. Nilsson (2003), A synthesis of the impact of Russian forests on the global carbon
budget for 1961-1998.
, Tellus, 55B, 391-415.
Suntharalingam, P., et al. (2005), The Influence of Reduced Carbon Emissions and Oxidation on
theDistribution of Atmospheric CO2: Implications for Inversion Analyses, Global Biogeo. Cycles, in press.
Takahashi, T., et al. (2002), Global sea-air CO2 flux based on climatological surface ocean pCO(2), and
seasonal biological and temperature effects, Deep-Sea Research Part Ii-Topical Studies in Oceanography,
49, 1601-1622.
Takahashi, T., et al. (1999), Net sea-air CO2 flux over the global oceans: An improved estimate based on the
sea-air pCO2 difference, paper presented at 2nd CO2 in Oceans Symposium, Tsukuba, Japan.
Tans, P. P., et al. (1995), Storage versus flux budgets: the terrestrial uptake of CO2 during the 1980s, in
Biotic feedbacks in the global system, edited by G. M. Woodwell and F. T. Mackensie, pp. 351-366, Oxford
University Press, New York.
Tans, P. P., et al. (1990), Observational Constraints On the Global Atmospheric Co2 Budget, Science, 247,
1431-1438.
Trumbore, S. E., et al. (1996), Rapid exchange between soil carbon and atmospheric carbon dioxide driven
by temperature change, Science, 272, 393-396.
van der Werf, G. R., et al. (2004), Continental-Scale Partitioning of Fire Emissions During the 1997 to 2001
El Nino/La Nina Period, Science, 303, 73-76.
Wang, Y. P., and G. L. McGregor (2003), Estimating regional terrestrial carbon fluxes for the Australian
continent using a multiple constraint approach II. The atmospheric constraint’ , Tellus, B55, 290-304.
WMO/GAW (2000), World Data Centre for Greenhouse Gases Data Summary WDCGG N°22, The Japan
Meteorological Agency, in cooperation with World Meteorological Organization.
Do Not Cite or Quote
116099703
7-12
2/12/2016
Total pages: 13
LA1 Extended Draft
Chapter 7
IPCC WG1 Fourth Assessment Report
Do Not Cite or Quote
7-13
2/12/2016
Total pages: 13
1
2
116099703